- Email: [email protected]
Design, Synthesis and Pharmacology of Cannabimimetic Indoles
Bioorgmic & Medicinal Chemistry L.etters. Vol. 4. No. 4. pp. X3-566.1594 Copyright 8 1994 Elsevicr Science L&l Printed in Great Britain. All rights rcsetved 0960-894x/94 smo+o.oo
0960-894X(94)EOOO9-4
Design, Synthesis and Pharmacology of Cannabimimetic Indoles
John W. Huffman* and Dong Dai Department of Chemistry, Clemson University, Clemson, SC 2%34-I%5
JJSA
Billy R. Martin and David R. Compton Department of Pharmacology and Toxicology, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298-0613, USA
Abstract: Molecular modeling has been employed to design a new group of cannabimimetic 1-alkyl-2methyl3-( 1-naphthoyl)indoles. Cannabinoid activity was evaluated in vivo in the mouse and in vitro by determining the binding to the cannabinoid receptor. Maximum activity was found for the I-butyl. pentyl and hexyl analogs. A rationalixation for the alignment of these indoles with traditional cannabinoids is presented. In the thirty years since Gaoni and Mechoulam described the isolation and elucidation of structure of Ag‘II-K (1. d-tetrahydrocannabinol)t been developed*
a comprehensive set of structure activity relationships for cannabinoids has
In the past few years, however, several nontraditional cannabinoids have been prepared,
which include infer alia, a series of 3arylcyclohexanols, Although structural similarities
such as CP-55,940 (2) and related compounds.3
between THC (1)and analog 2 are apparent, several cannabimimetic
aminoalkylindoles have been described recently which bear no obvious structural resemblence to other active cannabinoids4
The most active of these indoles is WIN-55.212 (3). which acts in vitro at the same binding
site as THC4 and shows in vivo activity in the mouse model which is comparable to THC.5 A three point receptor model for cannabinoids has been suggested.6 which has been modified by calculations which define a receptor essential volume for the cannabinoid
receptor. 7 Inherent in all approaches to the structural
requirements for cannabinoid activity is the presence of a phenolic hydmxyl group at C- 1, and a lipophilic side chain at C-3.*,6,7 Indole 3 does not contain a phenolic hydroxyl, nor is it obvious which portion of this molecule corresponds to the lipophilic side chain present in active cannabinoids described previously. To reconcile the structural features inherent in indole 3 with those of more traditional cannabinoids, and perhaps to develop other cannabimimetic indoles, a study was initiated using the program PCMode1.l The structure of THC was simplified by reducing the length of the side chain to four atoms, and the morpholine unit For THC the nonaromatic rings were assumed to be in the half
of indole 3 was changed to a dimethylamino.
chair conformation, and side chain conformations within 1 Kcal/mole of the apparent global minimum were considered. For indole 3 several rotational isomers about the bonds to the carbonyl carbon and the aromatic rings were investigated, and the minimum energy conformation of the dimethylamino (morpholmo) group was employed. Various atoms of the ring systems were compared and the relative alignments of the lipophilic side
563
J. W. HUFFMANet al,
564
chains and appropriate portions of indole 3 were compared visually. The best fit was found to be the alignment indicated in structures 1 and 3, in which the 1,2,3,9
and 10 carbons and phenolic oxygen of the cannabinoid
correspond with the corresponding carbons and ketonic oxygen of the indole (The lettering indicated on structures 1 and 3 indicates which atoms were aligned for comparison.).
When the structures were super-
imposed using this alignment, the side chain of the cannabinoids and the nitrogen substituent of the indole corresponded well with each other.
CH3
3 If the indole nitrogen in cannabimimetic
4 5 indoles corresponds to C-l’ of cannabinoids
and the
substituent on the nitrogen corresponds to the remaining atoms of the side chain, it is predicted that 1-alkyl-Z methyl-3-(1-naphthoyl)indoles such as 4 should show cannabinoid activity. Modeling studies identical to those described above were carried out using 1, again with a n-butyl substituent at C-3, and 4, R=n-C3H7. Excellent correspondence
between the cannabinoid
side chain and the nitrogen substituent of 4 was observed. A
corollary to this hypothesis is that the naphthoyl substituent, or an alternative group which occupies a similar region in space is essential to cannabinoid activity.5.7 To test this hypothesis a series of acyl indoles related to 4 containing various substituents on nitrogen as well as two indoles related to 5 possessing 3-acyl-2-methyl-l-(4morpholinoethyl) structures have been prepared and their pharmacology has been evaluated. The 3-acylindoles (R=CI-I3 and R=(CH&+CH) prepared by Friedel-Crafts acylation of 2-methyl- 1-(4-morpholinylethyl)~d~e.9~t~ 1) were prepared from 2-methyl-3-(1-naphthoyDindole9
were
Indole derivatives 4 (Table
and the appropriate alkyl halide or tosylate. For
primary groups alkylation was carried out using KOH in DMSO and the appropriate alkyl bromide. The 2heptyl derivatives were prepared from the corresponding tosylates with KH in DMSO. The unoptimized yields
Cannabimimetic indoles
565
with primary halides were from 38% to 54% ; however the 2-heptyl derivatives gave elimination as the major reaction path and the purified alkylation products were obtain& in only 17% to 19% yield All of the indoles were evaluated in virro by measuring their ability to displace [3Hl CP-55,940 (2) from its binding site in a membrane preparation as described by Compton er al.11 The KI values for the indole derivatives, CP-55,940 (2). Ag-THC (1)and WIN-55.212 (3) are presented in Table 1. Several of the indole derivatives were also evaluated in vivo in a mouse model of cannabimimetic activity.5*12 These evaluations include measures of spontaneous activity, andnociception, hypothermia, and catalepsy. It has been shown that an average of the EDso values for these four behavioral procedures correlate well with K$t and the in vivo data in Table 1 are presented as these averages. Data for compounds 1,2 and 3 are also included in Table 1. Table 1. Ph armacology of Indoles 4 and 5, and Cannabinoids 1,2 and 3. Entry
Compound
KI (nM)
in vivo (ED% pM/Kg)
125Ozt250
>257
1
4 R = 2-Phenylethyl
2
4 R = 2-Cyclohexylethyl
46f13
65.8
3
4 R = n-Propyl
164ti2
not det.
4
4 R = n-Butyl
22f1.5
not det.
5
4 R = n-Pentyl
9.5f4.5
2.62
6
4 R = n-Hexyl
48f13
14.0
7
4 R = n-Heptyl
>lO.OOO
~189
8
4 R = RS-2 Heptyl 4 R = R-2-Heptyl
33fll 81f41
35.6 21.5
4 R = S-ZHeptyl
72f32
not det.
9 10 11
SR-CH3
>lO.OOO
>350
5 R = (CH3)$J=CH 1 A9-THC
>lO,OOO
13
>103 4.7oa
14
2 CP-55,940
12
15
3 WIN-55,212
41a 0.9248
0.55c
24d
6.53b
. Ref.11. b. Ref. 5. c. Ref 12(b). d. Ref. 15. notda. = notdeteamined. As predicted by the hypothesis developed using PCModel those 3-acylindoles (5) which lack a naphthoyl substituent at C-3, were devoid of cannabmoid activity (entries 11, 12). However, in agreement with the hypothesis, the 1-alkylindoles in which the nitrogen substituent provides a chain of four to six carbons show both in vitro and in vivo activity comparable to both THC and WIN-55,212 (entries 4 -10, 13, 15). It is known that in traditional cannabiiids
activity normally requires a side chain of four to seven carbons.h
Since the
modeling studies indicate that the indole nitrogen corresponds to C-l’ of the cannabiioid, then these results are consistent with the theory of structural alignment proposed above. It is also interesting to note that the racemic 1-(2heptyl) analog and both enanriomers have the same (within experimental error) in vitro activity (entries 8 10). Also, the n-heptyl derivative (entry 7) is devoid of activity, although the next lower homolog is quite active. The n-heptyl compound would correspond to an octyl side chain on a traditional cannabinoid and consequently would be expected to have little, if any, activity.
J. W. HUFFMANet al.
566
In a recent publication, Semus and Martin have concluded, on the basis of modeling studies, that the phenolic hydroxyl group of cannabinoids hydrogen bonds to the receptor.13 However, in neither the indoles described in Table 1, nor WIN-%,212 can this occur. It appears quite probable that in both ~nnabi~~tic indoles, and possibly traditional cannabinoids,
that the receptor serves as a hydrogen bond donor to the
cannabimoid, similar to the mechanism prolxxd and evaluated by Reggio et aLI In suck,
modeling experiments have been employed to design a new group of c~~a~~c
which possess activity comparable to that of traditional cannabinoids.
indoles
These readily available compounds
demonstrate this activity both in vitro and in vivo, provide additional evidence that the cannabinoid receptor hydrogen bonds to the substrate, and suggest a possible structural alignment between cannabimimetic indoles and traditional cannabinoids. The work at Clemson was supported by grant DAO3590, and that at Virginia Acknowi~gements. CommonweaIth University by grant DAO3672, both from the National Institute on Drug Abuse, as well as by the Virginia Commonwealth Center on Drug Abuse (VCU). We thank Dr. Richard H. Wallace of the University of Alabama for assistance in obtaining high resolution mass spectral data. REFERENCESAND NOTES Gaoni, Y.; Mechoulam, R. J. Am. Chem. Sot. 1944,86, 1646. (a) Razdan, R. K. Pharmacol. Rev. 1986,38,75. (b) Mechoulam, R., &vane, W. A.,Glaser, R. Cannnbinoid Geometry and Biological Behavior. In MatijuanaKannabinoids: Neurobiology and Neurop&ysiology; Murphy, L; Bar&e, A.; CRC Press, Boca Raton 1992; pp I-33. 3. Johnson, M. R.; Melvin, L. S. The Discovery of Nonclassical C~~b~o~d Anal e&s. in Can~~i~ as Turmeric Agents, M~~~~, R.; CRC press, Boca Raton, K986, pp 121145. 4. D’Ambra, T. E.; Estep, K. G.; Bell, M. R.; Eissenstat, M. A.; Josef, K. A.; Ward, S. J.; Haycock, D. A.; Baizman, E. R.; Casiano, F. M.; Begli, N. C.; Cbippari, S. M.; @ego, J. D.; Kullnig, R. K.: Dalev. G. T. J. Med. Gem. 1992.35. 124. 5. ~mpto~,‘D. R.; Gold, L. H.; Ward, S: J.;‘Balster, R. L.; Martin, B. R. J. P~~oi. @I. Ther. 1992.263. 1118. 6. Dev&e, k. A.; Dysan, F. A. III; Johnsson, M. R.; Melvin, L. S.; Hewlett, A. C. Mol. Pharmacol.. 1988,34, 605. 7. Reggio, P. H.; Panu, A. M.; Miles, S. J. Med Chem. 1993,36, 1761. 8. PCModel is a modified MM2MMPl pmgram which inccnpora~ the MODEL graphical interface and which uermits the direct comtison of several structures. PCModel is marketed bvr Serena _ Softwam, ~l~~n~on, IN. Bell, M. R.; D’Ambm, T. E.; Kumar, V.; Eissanstat, M. A.; Herrmann, J. L.: Wet&, J. R.; Rosi, 9. D.; Philion, R. E.; Daum, S. J.; Hlasta, D. J.; Kullnig, R. K.; Ackerman, J. H.; Ha&rich, D. R.; Luttinger, D. A.; Baizman, E. R.; Miller, M. S.; Ward, S. J. J. Med. Chem. 1991,34, 1099. 10. All new compounds were characterized by 1H and 1% NMR and gave either acceptable analytical or HRMS data. Purity was established by TLC and 13CNMR. 11. Compton, D. R.; Rice, K. C.; De Costa, B. R.; Razdan, R. K.; Melvin, L. S.; Johnson, M-R.; Martin, B.R. J. Pharmacol. Exp. Ther. 1993,26.5,218. 12. (a) Martin, B. R.; Compton, D. R.; Little, P. J.; Martin, ‘I’.J.; Beardsley, P. M. Pharmacological Evaluation of Agonistic and Antagonistic Activity of Cannabinoids. In ,!imture Activiiy Relationships in Cannabinoids , Rapaka, R. S.; Makriyannis, A. NIDA Research Monograph 79, National Institute on Drug Abuse, Rockville, MD, 1987, pp 108-122. (b) Little, P. J.; Co ton, D. R.; Johnson, M. R.; Melvin, L. S.; Martin, B. R. J. Pharmacol. E.qx Ther. 1988,247, % 1 6. 13. Semus, S. F.; Martin, B. R. Life Sci. 1990,46, 1781. Reggio, P. H.; Seltzman, H. H.; Compton, I3. R.; Martin, B. R. Mol. P~~coi. 1990,38,854. ::: Jansen, E.M.; Haycock, D. A.; Ward, S. J.; Seybold, V. S.. Brain Res. 1992,X75,93. ::
(Received in USA 23 November 1993; accepted 17 December 1993)
0960-894X(94)EOOO9-4
Design, Synthesis and Pharmacology of Cannabimimetic Indoles
John W. Huffman* and Dong Dai Department of Chemistry, Clemson University, Clemson, SC 2%34-I%5
JJSA
Billy R. Martin and David R. Compton Department of Pharmacology and Toxicology, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298-0613, USA
Abstract: Molecular modeling has been employed to design a new group of cannabimimetic 1-alkyl-2methyl3-( 1-naphthoyl)indoles. Cannabinoid activity was evaluated in vivo in the mouse and in vitro by determining the binding to the cannabinoid receptor. Maximum activity was found for the I-butyl. pentyl and hexyl analogs. A rationalixation for the alignment of these indoles with traditional cannabinoids is presented. In the thirty years since Gaoni and Mechoulam described the isolation and elucidation of structure of Ag‘II-K (1. d-tetrahydrocannabinol)t been developed*
a comprehensive set of structure activity relationships for cannabinoids has
In the past few years, however, several nontraditional cannabinoids have been prepared,
which include infer alia, a series of 3arylcyclohexanols, Although structural similarities
such as CP-55,940 (2) and related compounds.3
between THC (1)and analog 2 are apparent, several cannabimimetic
aminoalkylindoles have been described recently which bear no obvious structural resemblence to other active cannabinoids4
The most active of these indoles is WIN-55.212 (3). which acts in vitro at the same binding
site as THC4 and shows in vivo activity in the mouse model which is comparable to THC.5 A three point receptor model for cannabinoids has been suggested.6 which has been modified by calculations which define a receptor essential volume for the cannabinoid
receptor. 7 Inherent in all approaches to the structural
requirements for cannabinoid activity is the presence of a phenolic hydmxyl group at C- 1, and a lipophilic side chain at C-3.*,6,7 Indole 3 does not contain a phenolic hydroxyl, nor is it obvious which portion of this molecule corresponds to the lipophilic side chain present in active cannabinoids described previously. To reconcile the structural features inherent in indole 3 with those of more traditional cannabinoids, and perhaps to develop other cannabimimetic indoles, a study was initiated using the program PCMode1.l The structure of THC was simplified by reducing the length of the side chain to four atoms, and the morpholine unit For THC the nonaromatic rings were assumed to be in the half
of indole 3 was changed to a dimethylamino.
chair conformation, and side chain conformations within 1 Kcal/mole of the apparent global minimum were considered. For indole 3 several rotational isomers about the bonds to the carbonyl carbon and the aromatic rings were investigated, and the minimum energy conformation of the dimethylamino (morpholmo) group was employed. Various atoms of the ring systems were compared and the relative alignments of the lipophilic side
563
J. W. HUFFMANet al,
564
chains and appropriate portions of indole 3 were compared visually. The best fit was found to be the alignment indicated in structures 1 and 3, in which the 1,2,3,9
and 10 carbons and phenolic oxygen of the cannabinoid
correspond with the corresponding carbons and ketonic oxygen of the indole (The lettering indicated on structures 1 and 3 indicates which atoms were aligned for comparison.).
When the structures were super-
imposed using this alignment, the side chain of the cannabinoids and the nitrogen substituent of the indole corresponded well with each other.
CH3
3 If the indole nitrogen in cannabimimetic
4 5 indoles corresponds to C-l’ of cannabinoids
and the
substituent on the nitrogen corresponds to the remaining atoms of the side chain, it is predicted that 1-alkyl-Z methyl-3-(1-naphthoyl)indoles such as 4 should show cannabinoid activity. Modeling studies identical to those described above were carried out using 1, again with a n-butyl substituent at C-3, and 4, R=n-C3H7. Excellent correspondence
between the cannabinoid
side chain and the nitrogen substituent of 4 was observed. A
corollary to this hypothesis is that the naphthoyl substituent, or an alternative group which occupies a similar region in space is essential to cannabinoid activity.5.7 To test this hypothesis a series of acyl indoles related to 4 containing various substituents on nitrogen as well as two indoles related to 5 possessing 3-acyl-2-methyl-l-(4morpholinoethyl) structures have been prepared and their pharmacology has been evaluated. The 3-acylindoles (R=CI-I3 and R=(CH&+CH) prepared by Friedel-Crafts acylation of 2-methyl- 1-(4-morpholinylethyl)~d~e.9~t~ 1) were prepared from 2-methyl-3-(1-naphthoyDindole9
were
Indole derivatives 4 (Table
and the appropriate alkyl halide or tosylate. For
primary groups alkylation was carried out using KOH in DMSO and the appropriate alkyl bromide. The 2heptyl derivatives were prepared from the corresponding tosylates with KH in DMSO. The unoptimized yields
Cannabimimetic indoles
565
with primary halides were from 38% to 54% ; however the 2-heptyl derivatives gave elimination as the major reaction path and the purified alkylation products were obtain& in only 17% to 19% yield All of the indoles were evaluated in virro by measuring their ability to displace [3Hl CP-55,940 (2) from its binding site in a membrane preparation as described by Compton er al.11 The KI values for the indole derivatives, CP-55,940 (2). Ag-THC (1)and WIN-55.212 (3) are presented in Table 1. Several of the indole derivatives were also evaluated in vivo in a mouse model of cannabimimetic activity.5*12 These evaluations include measures of spontaneous activity, andnociception, hypothermia, and catalepsy. It has been shown that an average of the EDso values for these four behavioral procedures correlate well with K$t and the in vivo data in Table 1 are presented as these averages. Data for compounds 1,2 and 3 are also included in Table 1. Table 1. Ph armacology of Indoles 4 and 5, and Cannabinoids 1,2 and 3. Entry
Compound
KI (nM)
in vivo (ED% pM/Kg)
125Ozt250
>257
1
4 R = 2-Phenylethyl
2
4 R = 2-Cyclohexylethyl
46f13
65.8
3
4 R = n-Propyl
164ti2
not det.
4
4 R = n-Butyl
22f1.5
not det.
5
4 R = n-Pentyl
9.5f4.5
2.62
6
4 R = n-Hexyl
48f13
14.0
7
4 R = n-Heptyl
>lO.OOO
~189
8
4 R = RS-2 Heptyl 4 R = R-2-Heptyl
33fll 81f41
35.6 21.5
4 R = S-ZHeptyl
72f32
not det.
9 10 11
SR-CH3
>lO.OOO
>350
5 R = (CH3)$J=CH 1 A9-THC
>lO,OOO
13
>103 4.7oa
14
2 CP-55,940
12
15
3 WIN-55,212
41a 0.9248
0.55c
24d
6.53b
. Ref.11. b. Ref. 5. c. Ref 12(b). d. Ref. 15. notda. = notdeteamined. As predicted by the hypothesis developed using PCModel those 3-acylindoles (5) which lack a naphthoyl substituent at C-3, were devoid of cannabmoid activity (entries 11, 12). However, in agreement with the hypothesis, the 1-alkylindoles in which the nitrogen substituent provides a chain of four to six carbons show both in vitro and in vivo activity comparable to both THC and WIN-55,212 (entries 4 -10, 13, 15). It is known that in traditional cannabiiids
activity normally requires a side chain of four to seven carbons.h
Since the
modeling studies indicate that the indole nitrogen corresponds to C-l’ of the cannabiioid, then these results are consistent with the theory of structural alignment proposed above. It is also interesting to note that the racemic 1-(2heptyl) analog and both enanriomers have the same (within experimental error) in vitro activity (entries 8 10). Also, the n-heptyl derivative (entry 7) is devoid of activity, although the next lower homolog is quite active. The n-heptyl compound would correspond to an octyl side chain on a traditional cannabinoid and consequently would be expected to have little, if any, activity.
J. W. HUFFMANet al.
566
In a recent publication, Semus and Martin have concluded, on the basis of modeling studies, that the phenolic hydroxyl group of cannabinoids hydrogen bonds to the receptor.13 However, in neither the indoles described in Table 1, nor WIN-%,212 can this occur. It appears quite probable that in both ~nnabi~~tic indoles, and possibly traditional cannabinoids,
that the receptor serves as a hydrogen bond donor to the
cannabimoid, similar to the mechanism prolxxd and evaluated by Reggio et aLI In suck,
modeling experiments have been employed to design a new group of c~~a~~c
which possess activity comparable to that of traditional cannabinoids.
indoles
These readily available compounds
demonstrate this activity both in vitro and in vivo, provide additional evidence that the cannabinoid receptor hydrogen bonds to the substrate, and suggest a possible structural alignment between cannabimimetic indoles and traditional cannabinoids. The work at Clemson was supported by grant DAO3590, and that at Virginia Acknowi~gements. CommonweaIth University by grant DAO3672, both from the National Institute on Drug Abuse, as well as by the Virginia Commonwealth Center on Drug Abuse (VCU). We thank Dr. Richard H. Wallace of the University of Alabama for assistance in obtaining high resolution mass spectral data. REFERENCESAND NOTES Gaoni, Y.; Mechoulam, R. J. Am. Chem. Sot. 1944,86, 1646. (a) Razdan, R. K. Pharmacol. Rev. 1986,38,75. (b) Mechoulam, R., &vane, W. A.,Glaser, R. Cannnbinoid Geometry and Biological Behavior. In MatijuanaKannabinoids: Neurobiology and Neurop&ysiology; Murphy, L; Bar&e, A.; CRC Press, Boca Raton 1992; pp I-33. 3. Johnson, M. R.; Melvin, L. S. The Discovery of Nonclassical C~~b~o~d Anal e&s. in Can~~i~ as Turmeric Agents, M~~~~, R.; CRC press, Boca Raton, K986, pp 121145. 4. D’Ambra, T. E.; Estep, K. G.; Bell, M. R.; Eissenstat, M. A.; Josef, K. A.; Ward, S. J.; Haycock, D. A.; Baizman, E. R.; Casiano, F. M.; Begli, N. C.; Cbippari, S. M.; @ego, J. D.; Kullnig, R. K.: Dalev. G. T. J. Med. Gem. 1992.35. 124. 5. ~mpto~,‘D. R.; Gold, L. H.; Ward, S: J.;‘Balster, R. L.; Martin, B. R. J. P~~oi. @I. Ther. 1992.263. 1118. 6. Dev&e, k. A.; Dysan, F. A. III; Johnsson, M. R.; Melvin, L. S.; Hewlett, A. C. Mol. Pharmacol.. 1988,34, 605. 7. Reggio, P. H.; Panu, A. M.; Miles, S. J. Med Chem. 1993,36, 1761. 8. PCModel is a modified MM2MMPl pmgram which inccnpora~ the MODEL graphical interface and which uermits the direct comtison of several structures. PCModel is marketed bvr Serena _ Softwam, ~l~~n~on, IN. Bell, M. R.; D’Ambm, T. E.; Kumar, V.; Eissanstat, M. A.; Herrmann, J. L.: Wet&, J. R.; Rosi, 9. D.; Philion, R. E.; Daum, S. J.; Hlasta, D. J.; Kullnig, R. K.; Ackerman, J. H.; Ha&rich, D. R.; Luttinger, D. A.; Baizman, E. R.; Miller, M. S.; Ward, S. J. J. Med. Chem. 1991,34, 1099. 10. All new compounds were characterized by 1H and 1% NMR and gave either acceptable analytical or HRMS data. Purity was established by TLC and 13CNMR. 11. Compton, D. R.; Rice, K. C.; De Costa, B. R.; Razdan, R. K.; Melvin, L. S.; Johnson, M-R.; Martin, B.R. J. Pharmacol. Exp. Ther. 1993,26.5,218. 12. (a) Martin, B. R.; Compton, D. R.; Little, P. J.; Martin, ‘I’.J.; Beardsley, P. M. Pharmacological Evaluation of Agonistic and Antagonistic Activity of Cannabinoids. In ,!imture Activiiy Relationships in Cannabinoids , Rapaka, R. S.; Makriyannis, A. NIDA Research Monograph 79, National Institute on Drug Abuse, Rockville, MD, 1987, pp 108-122. (b) Little, P. J.; Co ton, D. R.; Johnson, M. R.; Melvin, L. S.; Martin, B. R. J. Pharmacol. E.qx Ther. 1988,247, % 1 6. 13. Semus, S. F.; Martin, B. R. Life Sci. 1990,46, 1781. Reggio, P. H.; Seltzman, H. H.; Compton, I3. R.; Martin, B. R. Mol. P~~coi. 1990,38,854. ::: Jansen, E.M.; Haycock, D. A.; Ward, S. J.; Seybold, V. S.. Brain Res. 1992,X75,93. ::
(Received in USA 23 November 1993; accepted 17 December 1993)